The sources of error in previous methods are discussed, and a different method that minimizes them is presented. The theory is discussed with the objective of designing improved apparatus to make accurate routine height measurements. Measurements have been made of the height of the 5577 A nightglow layer obtaining an average height of 97 +_ 3 km. TH• PROBL•,M variations at angles high above the horizon.Although the experiment did not make exactly Measuremerits of the height of nightglow simultaneous measurements, the time difference has appealed to experimenters for 15 years, was small. This method has been repeated by but the methods, often incorrectly termed tri-others with good results that. were consistent angulation, have in most cases been unsatis-with stated accuracy. A disadvantage of the factory. In the light of present knowledge we method is that one must wait for the sluggish can see that the poor results stemmed from airglow to move or vary. the method of measurement. A type of expert-An uncertainty that disturbed early investiment reported by Chamberlain [1961] as spa-gators was the possibility that the nightglow tial triangulation Was that of Davis [1951] was irregular on a large vertical scale, i.e., who, observing from a single station through mountains and valleys that would not be obthe night and following the movements of lea-served alike from different stations. Present tures assumed stationary with respect to iner-knowledge of the thickness (•10 km) and the tial space, calculated the h•ight. Many 'trian-lateral size of patches (•-100'0 kin) makes this gulation' papers appeared using this idea. Two fear groundless. Indeed, it should not be overobservers obtained a height by assuming that looked that, if such gross structure existed, features observed by one in France retained measurements of the mean height would be their form until observed by the other in facilitated, provided one had a rough prior California eighthourslater. knowledge of the proportions. In this case,

No abstract